6.9   Optimal Management Zone Delineation

Investigators

Carl R. Dillon, Agricultural Economics, cdillon@ca.uky.edu
Scott A. Shearer, Biosystems and Agricultural Engineering, shearer@bae.uky.edu
Tom Mueller, Agronomy, mueller@uky.edu

Cooperators

Mike Ellis, Kentucky Producer and Precision Agriculture User, wdemike@iglou.com

Introduction

One of the most basic and perplexing questions involved in variable rate technology within precision agriculture is the delineation of the profit maximizing management zones which are treated separately with respect to input application.  While more accurate information regarding the optimal level of input (e.g., fertilizer) is desirable on a fine scale, the fixed costs per zone or grid (e.g., soil sampling) associated with this greater accuracy will not justify the additional costs at some point.  The decision of how to delineate optimal management zones or grid sizes represents a great opportunity for profit while at the same time presenting a daunting and complex problem that confuses researchers, extension specialists, industry leaders and producers alike.  Consequently, producers are left facing the difficult decision of how to delineate management zones without suitable guidance.  Alternatively, some producers desire to establish uniform grid size (as opposed to variably sized and shaped management zones) and face the question of the best grid size to use.  Although some informal standards regarding the 2.5 acre grid size is often used, decision tools are needed to provide a more robust management zone delineation procedure and uniform grid size level projections for farmers using variable rate application of inputs.  The focus of this proposed research is upon this very fundamental issue of precision agriculture that is so critical to adequate economic implementation of variable rate application regardless of the input (seed, fertilizer, pesticide).

            The most appropriate method of responding to this research question is through a multidisciplinary framework.  Specifically, a model embodying the decision-making framework of the producer will allow for proper analysis of these questions.  A model that allows for the objective of maximization of profits subject to the constraints a producer faces reflects the production environment faced by the farm manager.  Ultimately, a farmer’s decision is derived from the underlying economic consequences of the potential courses of action being considered. In turn, the economic consequences are determined by the underlying production responses.  Therefore, agricultural economic results drive a producer’s decisions while the physical agricultural relationships (e.g., agronomic, engineering) provide the foundation for the economic results.  Additionally, this basic framework can be expanded to incorporate consideration of risk to more completely ascertain economic sustainability.

            While this work will be primarily used to assist corn, soybean and wheat crop producers, the techniques developed here will be suitable for a broader audience.  Because this work provides a missing key element for properly assessing alternatives regarding variable rate input application, this research is especially relevant to all row crop producers who use variable rate technology.

Objectives

            The fundamental purpose of this research project is to assist crop producers by providing procedures and information that will assist them in making sustainable decisions concerning how to optimally delineate management zones.  The procedures developed here can be generalized to several production inputs (e.g., N, P and K).  The specific objectives include:

1.       Identify and develop various procedures that permit the determination of the specific optimal regions to be included within a management zone,

2.       Assess the economic performance of these procedures to compare their relative success to the profit maximizing management zone as well as communicate these results to the public and provide decision aids for the public’s use, and

3.       Perform sensitivity analysis to ascertain how optimal management zones change/respond to fluctuations in the economic decision-making environment including consideration of risk.

Background

Background information in the form of a review of literature can serve to establish a basic framework for the proposed study. Included in this will be a general discussion of economic studies pertaining to precision agriculture. Studies that examine variable rate technology, grid sampling and management zones are then discussed to complete the background information.

            The economic feasibility of precision agriculture is a common underlying question of producers considering its adoption.   While the literature regarding the economic issues in the area of precision agriculture is rich with numerous studies, they are broad based and display a substantial number of philosophical discussions rather than quantitative evaluation as is common with new technologies.  General philosophical discussions have ranged from historically descriptive (e.g.- Lowenberg-DeBoer; Sonka and Coaldrake) to examining the research opportunities and challenges of the future (e.g.-Weiss).   Lowenberg-DeBoer and Swinton conducted a review of the economics of precision agriculture, finding that economic feasibility is dependent upon several factors including many components of the underlying economic, agronomic and engineering environment.   Precision agriculture has been shown to be profitable, not profitable or inconclusive with mixed results, depending on the crop, inputs and conditions.

            In addition to these diverse precision agriculture economic studies, three specific areas are worthy of attention for this research: variable rate technology, grid sampling and management zones.  Variable rate technology research has included analysis of such components as nitrogen management (e.g., Thrikawala et al.; Babcock and Pautsch), lime application (e.g., Bongiovanni and Lowenberg-DeBoer) and spatial break-even variability assessment (English, Roberts and Mahajanashetti).  Studies rely predominantly upon the use of an assumed level of grid sampling, avoiding the issue of optimal grid size or management zone determination with few exceptions (Thrikawala, Weersink and Kachanoski).  While some economic research investigates grid sampling issues (e.g., Lenz; Rehm et al.), there is a void in the literature for sound economic models to address optimal grid size which is exceeded only by the apparent lack of economic analysis of the determination of optimal management zone delineation.

            The research proposed in this project lies at the very heart of precision agriculture. It will assist in the establishment of a fundamental framework that will permit analysis of a very germane and basic question currently plaguing the successful implementation of variable rate technology.  Specifically, how does a producer identify the optimal management zone?  The innovative model formulation proposed within this research project permits the appropriate economic analysis to be conducted for comparison to the less data intensive and more farmer friendly management zone delineation procedures presently being conceived and tested by others (such as delineation by soil properties or electrical conductivity).  Therefore, this research also provides complementary economic comparison among alternative procedures for management zone delineation. Additionally, the techniques include the economic assessment of another very important question: What is the optimal uniform grid size?  Thus, while some producers will use the variably sized and shaped management zones and others will use uniform sized and shaped grids, optimal determination of both can be handled with the model proposed.  Furthermore, this study will provide insights into the establishment of practical and simple decision rules.

Procedures

The objectives of this study will be accomplished in three steps.  The first and second steps deal with determination of relevant management zone size and the third step concerns determination of the robustness of the decision.  Each step is described in detail.

            Step 1:   Initial ideas for management zone delineation include the use of uniform grids of various sizes, soil type designation and soil property determination.  The use of electrical conductivity in the definition of appropriate zones of management, in conjunction with research by Mueller, provides some promising potential.  Delineation based on soil type, topsoil depth and other characteristics also merits consideration.  Further identification of strategies for management zone delineation will be completed through additional communications with experts in precision agriculture including research faculty, extension faculty, agribusiness industry representatives and producers.  More extensive literature review will also be undertaken.

            Step 2:   A mathematical programming model embodying the economic decision framework of a representative Kentucky crop producer will be formulated.  The objective function will be to maximize net farm returns above selected relevant costs.  Decision variables will include application rates of relevant nutrients by management zone as well as selection of the number, position and size of management zones using soil test information from each management zone.  The model will allow for the proper identification of the profit maximizing management zone for each individual input applied (as determined by the available database).   The best management zone for one input will not restrict derivation of the best management zone for another input.   Constraints modeled will include environmental impacts, land available, input purchases, commodity sales and management zone calculations.

            Data required include yield results by cell, soil test grid samples, fertilizer application rates by cell, commodity price, fertilizer prices, soil sample test cost and area per cell.  Appropriate procedures (such as biophysical simulation or statistical regression) will be used to develop production response functions.  Biophysical simulation can be used to estimate the underlying crop yield for a constant technology by altering production and management practices.  Alternatively, underlying production functions could be estimated assuming various forms from an actual data set if available from an actual producer that possesses a limited number of but sufficient quantity of observations.  Yield could be estimated as a function of the level of nutrient to be variably applied, selected soil properties, selected weather variables and the like. As required, appropriate testing (multicollinearity, autocorrelation, heteroskedasticity and examination of alternative production function forms) would be undertaken.  Supporting data  (yields, soil samples, etc.) from producers and colleagues will be requested as a preferred choice with biophysical simulation serving as a second alternative if such data is unavailable.

            The most appropriate results will be incorporated into the economic model to allow for projection of yield results for each management zone as based upon average fertilizer application.  The model will therefore determine the management zone that each individual cell should optimally be allocated within.  Given the cost of a soil test for each zone, additional costs for increasing the number of zones is weighed against the yield result differentials for added accuracy.  Thus, an assessment of the relative performance of alternative strategies for delineating management zones can be performed.

            Step 3:   The sensitivity of the net returns and the chosen optimal management zone to changes in the economic environment is investigated through alterations in the economic model.  Systematic increases and decreases (e.g., 5, 10, 15%) of selected economic data will be included in the model and the new optimal solution determined.  The economic components altered will include commodity price, fertilizer prices and cost of soil sampling.  The sensitivity of economic results associated with different attitudes towards production risk would also be appropriate.  E-V, or mean-variance, analysis is a widely used and accepted method for analyzing risk which could be used if supporting data is available.

Expected Benefits

The purpose of this study is to establish a fundamental framework which will permit analysis of a very germane and basic question currently plaguing the successful implementation of precision agricultural variable rate technology: How does a producer identify the optimal management zone? The innovative model formulation proposed within this research project permits the appropriate economic analysis to be conducted for comparison to the less data intensive and farmer friendly management zone delineation procedures presently being conceived and tested by others (such as delineation by soil properties or electrical conductivity). Therefore, this research also provides complementary economic comparison among these alternative tools for management zone delineation.  Additionally, the techniques include the economic assessment of another very important question: What is the optimal uniform grid size?  Furthermore, this research will provide insights into the establishment of practical and simple decision rules that may be used to this end by producers using precision agriculture.

Ultimately, this research aims at providing the missing element to permit economic comparison of alternative procedures for management zone delineation and for the improvement of these decision rules. Dissemination of the decision rules and their performance is anticipated through web pages, popular articles, extension materials, presentations and field days.

Deliverables

The many deliverables of this research include the establishment of an economic model that can be used in making economic comparison of proposed management zone delineation techniques. Actual comparison of alternative decision rules will be made for case studies.  Model results could provide insights useful in the refinement of current and developing management zone delineation rules. It is also hoped that model results lead to the development of actual decision rules if the need arises.  Refereed journal articles, popular articles including materials for web pages, presentations and extension materials including items suitable for field days and decision support tools are the physical deliverables anticipated from the project.  These different outlets are used in order to gain the greatest exposure to different audiences: other researchers and extension specialists, county agents, producers and others directly.  Additionally, it is noted that in today’s electronic age it is necessary to provide information via the Internet and such will be done through various web sites (e.g., University of Kentucky College of Agriculture).  Presentations, professional meetings, and in-service training with county agents will be provided to facilitate dissemination of information directly to producers.  Furthermore, communication of results through extension newsletters, extension publications and field days is envisioned.

While direct model usage is inappropriate for most producers and the underlying production functions are farmer and area dependent, the establishment of a framework of economic analysis represents the first stage of solving the problem of optimal management zone delineation. Furthermore, a means of evaluating practical, farmer oriented decision rules will therefore be available as well as establishing potential rules of thumb.